The North Atlantic realm is strongly affected by the Iceland–Jan Mayen mantle-hotspot system. Its strength has varied over time, with consequences for evolving plate-driving forces and topography maintained dynamically by convective stresses in the mantle. Here, we combined reconstructions of Atlantic spreading rates with maps of geological hiatuses at the scale of continental Europe. We extracted hiatus information from a geological map of Europe (scale 1:5 million)—at the available temporal resolution of geological series—to construct hiatus maps for the basal Paleocene to Pliocene Series boundaries across the North Atlantic realm and Europe. The maps allow one to visualize first-order changes in hiatus intensity and spatial extent; they serve as a reference frame and proxy tool for paleogeographic analysis of topography; and they reveal that episodes of spreading acceleration in the North Atlantic coincide with episodes of erosion/nondeposition intensification at the continental scale of Europe. Such patterns are not predicted by the plate mode of mantle convection, but they are generally consistent with the plume mode. Interregional-scale hiatus mapping may serve as a powerful new technique to visualize these effects, especially once the temporal resolution of the maps has been refined. Interpretation of such hiatal surfaces is most effective in a context with theoretical frameworks, such as the plume mode of mantle convection.
Flow in Earth’s mantle induces both horizontal plate motion (Davies, 1999) and vertical deflections of Earth’s surface (e.g., Pekeris, 1935). The latter were inferred early on from the sedimentary record through unconformities (e.g., Stille, 1924; Rainbird and Ernst, 2001). Termed “dynamic topography” by Hager et al. (1985), they are considered in models of the geoid, because the mass anomalies associated with surface deflections yield gravity anomalies of comparable amplitude to the flow-inducing mantle density variations. Geoid interpretations performed with dynamic Earth models therefore account for dynamic topography as well as mantle density heterogeneity (e.g., Richards and Hager, 1984). The dynamic topography response of Earth models to internal loads (e.g., hot rising plumes or cold sinking slabs) is expressed through kernels (see Colli et al., 2016). For a plume rising through a uniform-viscosity mantle, they predict that deflections will grow continuously during plume ascent, whereas in the presence of a weak upper mantle, i.e., an asthenosphere, much of the surface deflection develops in the final phase of plume ascent. This makes rapid surface uplift events in a time span of a few million years geodynamically plausible.
Geodynamicists distinguish between the so-called plate and plume modes of mantle convection (e.g., Davies, 1988a, 1988b). The former relates to descent of oceanic lithosphere into the mantle by subduction, and the latter relates to the rise of plumes from a hot thermal boundary at the core-mantle boundary. Above subducting slabs, dynamic topography induced by the plate mode is large in spatial scale (∼10,000 km) and changes slowly over time (Mueller et al., 2018). Its large spatial scale and slow temporal change were suggested by Chase (1979), who noted the correspondence between major geoid lows and Cretaceous subduction, implying that dynamic topography related to the plate mode evolves on time scales of ∼100 m.y. Evidence from passive continental margins points to shorter length scales and notably faster vertical motion, on the order of 10 m.y., with episodes of regional surface uplift and subsidence seemingly unrelated to large-scale mantle flow (e.g., Green et al., 2018). These episodes link to the plume mode.
Interest in the vertical motion of the lithosphere as a probe into mantle flow is growing (e.g., Bunge and Glasmacher, 2018). Following Morgan et al. (1995), who connected plumes and upper-mantle flow through a plume-fed asthenosphere, the series of papers by Höink and Lenardic (2008, 2010) and Höink et al. (2011, 2012) laid the foundation to explore mantle convection in the context of so-called Poiseuille/Couette flow. This flow type increases lithosphere-asthenosphere coupling (Höink et al., 2012) and yields rapid horizontal asthenosphere flow velocities of more than 10 cm/yr (Weismueller et al., 2015). The latter provides active driving shear beneath the lithosphere. Importantly, Poiseuille/Couette flow relates nonisostatic vertical motion explicitly to basal shear stress variations below the lithosphere. It thus connects spreading-rate changes and coeval episodes of epeirogenic motion, as documented in the South Atlantic region (e.g., Colli et al., 2014), through a geodynamically plausible model.
The North Atlantic realm is well suited to explore the temporal and spatial links between changes in spreading rates and dynamic topography related to mantle upwelling events. The region has long been shielded from subduction and is located above the Iceland–Jan Mayen plume system, which is well imaged seismically (e.g., Rickers et al., 2013; Schaeffer and Lebedev, 2013). The region also has well-preserved seafloor magnetic lineations (Pitman et al., 1971; Emery and Uchupi, 1984, and references therein; Mueller et al., 2008a; Hopper et al., 2014, and references therein, their fig. 5.1), which have been utilized to construct regional and global plate-motion models (Gaina et al., 2002; Seton et al., 2012). In adjacent continental regions, Paleogene and Neogene uplift episodes are well documented in the British Isles, Greenland/Labrador, and Scandinavia (Riis, 1996; Dam et al., 1998; Lidmar-Bergström and Näslund, 2002; Japsen et al., 2005; Praeg et al., 2005; Hillis et al., 2008; Anell et al., 2009), and these are also supported by sediment compaction studies (Japsen, 2018) and thermochronologic (Zattin et al., 2016) and stratigraphic observations. The latter are summarized in the Stratagem project (Stoker and Shannon, 2005, and references therein), by Lundin and Doré (2002), and in a review by Anell et al. (2012). Accounts also exist of repeated burial and exhumation (e.g., Japsen et al., 2012) and transient surface uplifts (Hartley et al., 2011; Lovell, 2010) with amplitudes approaching 1 km.
Comparison of integrated geological observations against geodynamic models often is hampered by the lack of internally consistent spatial and temporal reference frames. Some studies are based on sufficiently large spatial frames, but detailed temporal information is confined to sectional profiles or chronostratigraphic charts that are not spatially connected (e.g., Dam et al., 1998; Japsen and Chalmers, 2000; Praeg et al., 2005; Bonow et al., 2007; Anell et al., 2009). If they are connected, these frames typically are truncated laterally before the signals of interest diminish to zero (e.g., cf. figs 37 and 43 inGreen et al., 2013; chapter 7.1 inHopper et al., 2014, in particular stratigraphic correlation panels in their figs. 7.14–7.20). In this regard, geological maps provide a unique opportunity to connect geological units laterally to show the transient and interlocking patterns of vertical motion of Earth’s surface as recorded within sedimentary basins and through unconformities (e.g., Levorsen, 1933; Sloss et al., 1949; Wheeler, 1958; Friedrich, 2019). To accommodate the growing number and diversity of high-resolution data, most maps are confined to local or regional scales, but continental-scale and world maps (at the temporal resolution of systems or series) have been standardized by international efforts (International Commission of the Geological Map of the World [ICGMW]). Because the scale of mantle plume–related processes requires the synthesis of geological data at equally large dimensions, i.e., on so-called interregional or continental scales, maps standardized through the ICGMW are ideally suited to integrate geological observations at the temporal resolution currently confined to chronostratigraphic charts (e.g., Hopper et al., 2014) and compare them against geodynamic models.
The purpose of our study was to visualize information about the long-wavelength vertical motion of Earth’s surface contained within the 1:5,000,000 scale International Geological Map of Europe and Adjacent Regions (IGME 5000; Asch, 2003, 2005) and connect it to the spreading-rate changes of the northern Atlantic Ocean. In contrast to most studies, which are based on information gained from basins, we exploited the information from uplifted and eroded regions, which are preserved in the geological record as unconformities and are shown on maps as hiatuses (time missing from the rock record), further developing the paleogeological hiatus mapping technique of Friedrich (2019). This approach allows us to map the dimensions and shapes of hiatus areas for each major base datum, but it is limited by the temporal information in the underlying map.
North Atlantic Spreading Rate History
The North Atlantic seafloor magnetic record (Fig. 1A) shows that the basin has undergone significant spreading rate changes. The scatter plot in Figure 1B makes them evident by comparing the width of seafloor magnetic lineations to their duration. At a constant spreading rate, the relationship between the isochron’s width and its duration should be linear (Fig. 1B). However, the numerous outliers in Figure 1B, which represent time intervals during which isochrones formed anomalously fast or slow, indicate that this is not the case. They show two main episodes of spreading rate changes: in the Paleocene–Eocene and the Miocene. The rhombi falling below the trend line correspond to ages of 60–50 Ma and 20–10 Ma, when spreading was fast. Conversely, the outliers above the trend correspond to post–10 Ma, 40–20 Ma, and pre–60 Ma time, when spreading was slow. The North Atlantic spreading changes are even more clear in map view (Fig. 2), based on a quaternion-based spreading velocity code (Clark, 2018) and rotational models for the Paleogene (Gaina et al., 2002) and the Neogene (Merkouriev and DeMets, 2014). High-resolution rotational data sets, like the one reported by Merkouriev and DeMets (2014), tend to be noisy. Because this affects the spreading velocity calculations, we employed the Redback open-source noise-reduction software based on Bayesian inference to find best-fitting solutions to stage rotations (Iaffaldano et al., 2014) for the Neogene data. Velocities were computed at 1 m.y. intervals from 60 Ma to the present. The spreading rate was calculated for seven points along the mid-oceanic ridge in 5° latitude increments. As expected, spreading rates are higher at lower latitudes, where the distance to the Euler pole is larger. The rates were faster between 60 and 40 Ma and between 20 and 10 Ma, separated by slower spreading velocities between 40 and 20 Ma and since 10 Ma, as seen from Figure 1.
Interregional Paleogeological Hiatus Surface Maps
As noted above, numerous Cenozoic uplift and exhumation events have occurred across the North Atlantic region (for reviews, see Japsen and Chalmers, 2000; Anell et al., 2009; Hopper et al., 2014). Because it is difficult to integrate local and heterogeneous data from basins and mountains into a global reference frame for vertical motion, we exploited information contained within standard interregional-scale geological maps as a new interregional-scale reference frame for vertical surface motion. We constrained hiatal surfaces captured at the base datum of geological systems or series using a method introduced by Friedrich (2019) and Friedrich et al. (2018). Here, we also went one step further and estimated the shapes and dimensions of uplifting regions at the continental scale of Europe based on interpolation techniques applied to the digital version of the IGME 5000 (Asch, 2005).
In a paleotectonic reference frame (Mueller et al., 2008a), we extracted the numerous nonconformable geological contacts on the 1:5,000,000 scale International Geological Map of Europe and Adjacent Areas (Asch, 2005), which has a temporal resolution of series, i.e., tens to a few tens of millions of years. We then constructed interregional hiatal surface maps for the base datum of the Paleocene to Pliocene Series, respectively, by mapping hiatus duration—the gap in geological time at map resolution—relative to underlying (older) series. If no series is absent, the contact is conformable at the mapped resolution (hiatus = 0 m.y.), but if at least one is absent, the contact is unconformable and represents a time interval of significant nondeposition and/or erosion (hiatus > 0 m.y.; Figs. 3A and 3B). Equal hiatus values, which reflect the cumulative history of erosion and nondeposition at any location, were then contoured across the map, as shown schematically in Figure 3C. The rationale for extrapolating hiatus areas across the map is the assertion that any pixel on the map represents either a status of uplift and erosion or subsidence and sedimentation on the interregional scale at any given time. Of course, later calibration using additional data and a refined temporal resolution of the base map will help to define the hiatus area maps more rigorously and reduce temporal uncertainty.
We then treated hiatus surfaces as any other geological contact surfaces that are subject to erosion by renewed uplift. In such cases, it is permissible that a hiatus surface may be eroded in regions that experienced renewed uplift, as shown in Figure 4. Across our study region, large areas of southern Greenland, the Scandes, and the British Isles may have been affected by such processes. The full dimensions and shapes of the uplift and erosional exhumation are the subjects of this paper and further studies.
We represent the permissible spatial extent of each hiatus in Figure 5 and also show the geologic contacts extracted from the IGME 5000 (Asch, 2003, 2005) for comparison in Figure 6. To construct the paleogeological hiatus maps, we used a nearest-neighbor interpolation to map hiatus intensity and the spatial extent of each hiatus in Figure 5. The algorithm assigns to each unknown node a value equal to the weighted average of known points in a 20 km search radius, with the weight accounting for the distance to the known points. A new node value is assigned only if there are at least four defined node values in the node’s search radius. Otherwise, the algorithm outputs an undefined solution. We found that the interpolation method defines new node values only based on sufficient input data and that it avoids unrealistic curvature resulting from polynomial or spline interpolation.
The hiatus maps for the basal boundaries of the Paleocene to Pliocene Series across the North Atlantic realm and Europe reveal first-order differences in hiatus surface intensity, spatial dimension, and location (Figs. 5A–5E). The Paleocene basal hiatal surface is limited to the northern North Atlantic breakup area, whereas the Eocene basal hiatal surface coherently covers most of western, southern, and eastern Europe (Figs. 5A and 5B). In contrast, the Oligocene basal hiatal surface is restricted to Iberia and central Europe (Fig. 5C). The Miocene basal hiatal surface dominates in NW Africa, SW Europe, eastern Europe, and across the Russian Platform (Fig. 5D). The Pliocene basal hiatal surface occurs on the British Isles, France, Scandinavia, and from the Barents Sea to the Caspian Sea (Fig. 5E). The respective hiatal surfaces areas, which were constructed based on the data shown in Figure 6, are plotted in Figure 7, and these were compared with the temporal variations in spreading rate across the North Atlantic seafloor discussed above. Blue regions in Figures 5A to 5E represent areas of no hiatus within the mapping resolution. They are of a qualitative character based on the interpolation described above.
Uncertainties and Limitations
Currently, the paleogeological hiatus mapping method is subject to a number of limitations and uncertainties. We note that the hiatus duration displayed on the IGME 5000 (Asch, 2003, 2005) may be shorter than the real local hiatus duration (Friedrich, 2019). This effect may be corrected once the information available on local chronostratigraphic charts is transferred to the interregional-scale maps. The automatic interpolation routine depicts hiatal surfaces displayed on the IGME 5000 regardless of their geological origin.
The time interval represented by an interregional hiatus is spatially diachronous (cf. Sengör, 2016; Friedrich et al., 2018) and significantly larger than the interval during which the hiatus-forming process(es) took place. The temporal resolution of currently available digital continent-scale geological maps is tens of millions of years (systems) to a few tens of millions of years (series), whereas geological processes of interest require resolution of 1–2 m.y., i.e., stages. Therefore, our analysis carries uncertainties affecting the spatial distribution, intensity, and temporal resolution of hiatus surfaces, which we must state clearly: (1) The interpolation choice affects the spatial extent of hiatal surfaces, as well as hiatus intensity estimates. (2) Very large hiatus values, which imply a strong erosion/uplift episode at a given time, may have resulted from earlier epeirogeny or orogeny. (3) Unconformities might have been eroded (e.g., Fig. 4) or covered by younger sedimentary successions and are thus not visible by our automatic mapping method. (4) Hiatus area does not show a steady increase with younger age, which could be a potential source of uncertainty, as younger strata may cover potential hiatus areas. In spite of these uncertainties, our results agree with a range of geologic data (see GSA Data Repository Table DR21 for a summary).
Comparison of Paleogeological Hiatus Areas and Spreading Rate Changes
Our hiatus plotting method provides insight into the paleotopography by revealing areas of erosion or nondeposition and areas of sedimentation captured at distinct stratigraphic boundaries. For the purpose of our study, variations in relative elevation and paleotopography serve as indicators of possible surface vertical motions. Our maps of hiatus surfaces reveal significant interregional temporal changes (Fig. 5). Hiatus surfaces abound in the Paleocene and Eocene, diminish in the Oligocene, and regain areal extent during the Miocene and Pliocene (Fig. 7). High spreading rates during Paleocene–Eocene time and the Miocene are interrupted by slower spreading in the Oligocene, akin to the temporal changes of hiatal surface area.
The spreading rates and the hiatus surface areas (Fig. 7) vary on short time scales relative to a mantle overturn (Bunge et al., 1998). This makes it attractive to invoke plate-boundary forces, because the latter can vary rapidly compared to variations in the large-scale mantle circulation pattern (Iaffaldano and Bunge, 2009). Gaina et al. (2009) argued that evolving plate-boundary forces associated with the formation of the Jan Mayen microcontinent caused the North Atlantic spreading changes, but it is difficult to see how this process would induce hiatus surface changes as far away as the Russian Platform. Changes in hiatus surface in turn could reflect long-term sea-level variations (Mueller et al., 2008b), but this leaves the coeval spreading rate changes unexplained. Importantly, coeval evolution of spreading rates and vertical motion in the surrounding continents has been reported for the less complex South Atlantic spreading system (Colli et al., 2014), where it has been attributed to pressure-driven flow in an asthenosphere layer (Colli et al., 2013).
Our maps show that hiatus surface area at the continental scale of Europe is characterized by length scales of 1000 km to a few thousand kilometers. The scale compares favorably to the wavelength inferred from residual depth measurements for present-day dynamic topography in the oceanic realm (Hoggard et al., 2017). The complex architecture of continental lithosphere makes estimates of present-day onshore dynamic topography more difficult. However, in the case of Africa (Guillocheau et al., 2018), geomorphic arguments using planation surfaces were applied to suggest that present-day central Africa topography records long-wavelength deformations of ∼1000–2000 km, manifested by the growth of the Cameroon and East African Dome. A similar dimension for landscape evolution was reported for the central Atlantic passive margin from low-temperature thermochronology by Sehrt et al. (2018), who argued that recent surface uplift initiated in the Miocene. Such a pattern is not predicted by the plate mode of mantle convection, but it is generally consistent with the plume mode.
A sublithospheric origin for coeval changes in spreading rate and hiatus surface area in the North Atlantic realm is geodynamically plausible. Höink et al. (2011) used scaling arguments to predict that plate motions in the Atlantic region are strongly driven by basal shear from asthenosphere flow, sourced from hotspots in the region. Near the western European margin, these include the Canary Islands (Troll and Carracedo, 2016), which experienced substantial Miocene uplift at the local (Meco et al., 2007) and regional scale (Sehrt et al., 2018), and the prominent Iceland–Jan Mayen plume system. The broad extent of the latter was imaged recently with full waveform inversion by Rickers et al. (2013).
It is likely that plume flux in the North Atlantic region varied over time. There are reports on mass flux (Poore et al., 2009) and thermal (Spice et al., 2016) variations for the Iceland–Jan Mayen plume system, with peaks in the early and late Cenozoic and a minimum in the Oligocene. Time-dependent upper-mantle flow has also been invoked to explain transient landscapes in the North Atlantic, now buried beneath marine sediments, inferring the passage of sublithospheric material at velocities in excess of 20 cm/yr (Hartley et al., 2011). While uplift in the North Atlantic realm has been linked to the arrival of the Iceland plume before (for a review, see Saunders et al., 2007), our results suggest substantial far-field effects of plume flux variations across Europe and the adjacent areas throughout the Cenozoic, transmitted presumably through pressure-driven upper-mantle flow (see Fig. 8).
Inverse theory based on adjoint equations (e.g., Bunge et al., 2003; Ismail-Zadeh et al., 2004) allows geodynamicists to reconstruct mantle flow back in time. This allows us to test poorly known geodynamic modeling parameters of mantle flow with interregional-scale geological observations, for instance, through predictions of mantle-induced past dynamic topography (Colli et al., 2018), which we report in Figure 7. While uncalibrated maps of hiatus surface area mask substantial temporal uncertainty, their use as indicators of evolving dynamic topography in connection to oceanic spreading variations has the potential to provide key geologic constraints on past mantle flow.
Estimates of NW European hiatus distribution indicate that episodes of erosion/nondeposition intensification coincided with episodes of North Atlantic spreading acceleration. Although the underlying mechanisms for these observations are still poorly understood, it is likely they reflect dynamic mantle processes, indicating episodes of invigorated asthenosphere flow through pressure-driven flow, which links horizontal and vertical plate motions. While emphasizing the large uncertainties of our approach, we suggest that interregional hiatus mapping as an indicator of evolving mantle-induced dynamic topography should be extended to other continents so that the link to plate-motion variations and sublithospheric flow can be tested further. Compilation of interregional geological maps at the resolution of stages (1–2 m.y.) and calibration with geological observations from sedimentary basins and eroding regions will further reduce spatial and temporal uncertainty.
We thank the German Research Foundation (DFG) South Atlantic Margin Processes and Links with onshore Evolution (SAMPLE) Priority Program for support, Lithosphere Science Editor Kurt Stüwe and reviewer Paul Ryan for constructive comments, and Peter Japsen for discussions relating to local aspects addressed in this paper. The research was supported by the Norwegian Research Council as part of the Industrial Ph.D. program under grant number 238790/O30.